On Wed, Oct 12, 2016 at 10:15:52PM -0400, Theodore Ts'o wrote:
> On Wed, Oct 12, 2016 at 03:19:25PM -0600, Jens Axboe wrote:
> >
> > FWIW, this is the commit that fixes it:
> >
> > commit 39d13e67ef1f4b327c68431f8daf033a03920117
> > Author: Jens Axboe <axboe@xxxxxx>
> > Date: Fri Aug 26 14:39:30 2016 -0600
> >
> > backend: check if we need to update rusage stats, if stat_mutex is busy
> >
> > 2.14 and newer should not have the problem, but earlier versions may
> > depending on how old...
>
> Unfortunately, I'm still seeing hangs in generic/299 with the latest version of fio:
>
> fio fio-2.14-10-g0a301e9 (Fri, 23 Sep 2016 11:57:00 -0600)
>
> If I use an older fio, it reliably does not hang. What can I do to
> help debug this?
>
> As I said, I can attach to the hanging fio using a gdb and give you
> stackdumps for all of the threads if that would be helpful.
I don't have git fio hangs on 299 (just a lot of "fallocate: No space left on
device" in results/generic/299.full):
# fio -v
fio-2.14-27-gafd2c
# mount tmpfs -t tmpfs -o size=26g /ramdisk/
# fallocate -l 12g /ramdisk/testvol1 && losetup -f /ramdisk/testvol1
# fallocate -l 12g /ramdisk/scratch0 && losetup -f /ramdisk/scratch0
root@ttip:/home/mator/xfstests-dev# cat local.config
export TEST_DEV=/dev/loop0
export TEST_DIR=/testvol
export SCRATCH_DEV_POOL="/dev/loop1"
#export SCRATCH_DEV_POOL="/dev/loop0 /dev/loop1 /dev/loop2 /dev/loop3"
export SCRATCH_MNT=/mnt/scratch
# mkfs.xfs -f /dev/loop0
meta-data=/dev/loop0 isize=512 agcount=4, agsize=786432 blks
= sectsz=512 attr=2, projid32bit=1
= crc=1 finobt=1, sparse=0, rmapbt=0
data = bsize=4096 blocks=3145728, imaxpct=25
= sunit=0 swidth=0 blks
naming =version 2 bsize=4096 ascii-ci=0 ftype=1
log =internal log bsize=4096 blocks=2560, version=2
= sectsz=512 sunit=0 blks, lazy-count=1
realtime =none extsz=4096 blocks=0, rtextents=0
root@ttip:/home/mator/xfstests-dev# ./check generic/299
FSTYP -- xfs (non-debug)
PLATFORM -- Linux/sparc64 ttip 4.8.0+
MKFS_OPTIONS -- -f -bsize=4096 /dev/loop1
MOUNT_OPTIONS -- /dev/loop1 /mnt/scratch
generic/299 103s
Ran: generic/299
Passed all 1 tests
kernel logs:
$ journalctl -k -e
Oct 13 15:35:41 ttip unknown: run fstests generic/299 at 2016-10-13 15:35:41
Oct 13 15:35:41 ttip kernel: XFS (loop1): Unmounting Filesystem
Oct 13 15:35:42 ttip kernel: XFS (loop1): Mounting V5 Filesystem
Oct 13 15:35:42 ttip kernel: XFS (loop1): Ending clean mount
Oct 13 15:37:24 ttip kernel: XFS (loop1): Unmounting Filesystem
Oct 13 15:37:25 ttip kernel: XFS (loop0): Unmounting Filesystem
results/generic/299.full in attachment
Attachment:
299.full.gz
Description: GNU Zip compressed data
